A deep learning library for spiking neural networks which is based on PyTorch, focuses on fast training and supports inference on neuromorphic hardware.
For now we use the tensor with shape [Batch * Time, polarity, H, W] which is kind of ambigous for new users of Sinabs.
To compatible with nn.Conv2d, we usually use IAFSqueeze, which always need to care about the batch_size argument. If we use nn.Conv3d the only thing we need to care is whether the Time of input data is equal to each other.
Description
For now we use the tensor with shape
[Batch * Time, polarity, H, W]which is kind of ambigous for new users of Sinabs.To compatible with
nn.Conv2d, we usually useIAFSqueeze, which always need to care about thebatch_sizeargument. If we usenn.Conv3dthe only thing we need to care is whether theTimeof input data is equal to each other.